并非所有线索都能被发现：事实分布和“不要编造”提示如何影响长上下文语言模型中的检索、推理和幻觉
Not All Needles Are Found: How Fact Distribution and Don't Make It Up Prompts Shape Retrieval, Reasoning, and Hallucination in Long-Context LLMs
AI总结：
研究大语言模型中事实分布和反幻觉提示对检索、推理及幻觉的影响，通过扩展基准评估多个模型，识别出分布崩溃和安全代价两种失败模式，发现许多失败源于无效上下文利用，强调特定模型稳健性和上下文管理的重要性。
摘要：
随着大语言模型（LLMs）越来越多地使用大规模上下文窗口作为自主任务的工作记忆，其可靠性会因信息在真实语料库中的分布方式而显著波动。我们通过引入一个为可扩展性设计的、与模型无关的扩展大海捞针基准来研究事实放置、语料库级分布和反幻觉（“不要编造”）提示如何影响模型行为，并将其应用于评估Gemini-2.5-flash、ChatGPT-5-mini、Claude-4.5-haiku和Deepseek-v3.2-chat。与先前工作不同，我们分别评估字面提取、逻辑推理和幻觉风险。我们识别出两种关键的失败模式：分布崩溃，即证据分散时性能显著下降；以及安全代价，即反幻觉提示导致对现有事实和证据过度保守地拒绝，大幅降低准确性。我们的结果表明，许多失败源于无效的上下文利用，因为即使相关信息存在，模型也难以对其进行优先排序。这些发现凸显了针对特定模型的稳健性和有效的上下文管理的必要性，以确保在长期代理工作流程中的可靠部署。
英文摘要：
As Large Language Models (LLMs) increasingly utilize massive context windows as working memory for autonomous tasks, their reliability fluctuates significantly depending on how information is distributed in real-world corpora. We investigate how fact placement, corpus-level distributions, and anti-hallucination ("Don't Make It Up") prompts influence model behavior by introducing a model-agnostic extended needle-in-a-haystack benchmark designed for scalability, which we apply to evaluate Gemini-2.5-flash, ChatGPT-5-mini, Claude-4.5-haiku, and Deepseek-v3.2-chat. Unlike prior work, we separately evaluate literal extraction, logical inference, and hallucination risk. We identify two critical failure modes: Distributional Collapse, where performance degrades significantly when evidence is dispersed; and a Safety Tax, where anti-hallucination prompts cause over-conservative refusal of present facts and evidence, sharply reducing accuracy. Our results suggest that many failures stem from ineffective context utilization, as models struggle to prioritize relevant information even when it is present. These findings highlight the need for model-specific robustness and effective context management to ensure reliable deployment in long-horizon agentic workflows.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
